Symptoms and Immune Markers in Plasmodium/Dengue Virus Co-infection Compared with Mono-infection with Either in Peru

Background Malaria and dengue are two of the most common vector-borne diseases in the world, but co-infection is rarely described, and immunologic comparisons of co-infection with mono-infection are lacking. Methodology and Principal Findings We collected symptom histories and blood specimens from subjects in a febrile illness surveillance study conducted in Iquitos and Puerto Maldonado, Peru, between 2002–2011. Nineteen symptoms and 18 immune markers at presentation were compared among those with co-infection with Plasmodium/dengue virus (DENV), Plasmodium mono-infection, and DENV mono-infection. Seventeen subjects were identified as having Plasmodium/DENV co-infection and were retrospectively matched with 51 DENV mono-infected and 44 Plasmodium mono-infected subjects. Those with Plasmodium mono-infection had higher levels of IL-4, IL-6, IL-10, IL-12, IL-13, IL-17A, IFN-γ, and MIP1-α/CCL3 compared with DENV mono-infection or co-infection; those with Plasmodium mono-infection had more cough than those with DENV mono-infection. Subjects with DENV mono-infection had higher levels of TGF-β1 and more myalgia than those with Plasmodium mono-infection. No symptom was more common and no immune marker level was higher in the co-infected group, which had similar findings to the DENV mono-infected subjects. Conclusions/Significance Compared with mono-infection with either pathogen, Plasmodium/DENV co-infection was not associated with worse disease and resembled DENV mono-infection in both symptom frequency and immune marker level.


Introduction
Malaria and dengue fever are two of the most important mosquito-borne infections affecting humans. Annually, approximately 584,000 and 12,000 people die and another 198 million and 96 million are estimated to fall ill from malaria and dengue, respectively [1,2]. Although Anopheles, the mosquito vector of malaria, and Aedes, the mosquito vector of dengue virus (DENV), occupy different ecological niches, considerable geographical overlap in disease risk exists for the 3.3 and 4.0 billion people who live in an area endemic for malaria and dengue, respectively [2,3]. The two diseases also share many common clinical features-including fever, headache, body aches, and fatigue-making one disease easily mistaken for the other.
Limited availability of diagnostics may be a reason only a handful of reports describe infection with Plasmodium species (which will be abbreviated Plasmodium) and DENV each year (Table 1)  . The first reports of co-infection occurred in 2005 [4,6] and most subsequent reports describe only individual patients and rely on serological methods for diagnosing DENV infection. The majority originate from India and nearby Pakistan, although West Africa [6], French Guiana [11,17], Brazil [19,23,27], Malaysia [25], Bangladesh [33,34], East Timor [9], Thailand [26], and Indonesia [24] have also been the origin of case reports.
The immune responses against DENV and Plasmodium parasites are unique and complex, although similarities between the two responses exist [35,36]. Both pathogens elicit a robust antibody response with the ability to neutralize receptor-mediated invasion of host cells [36,37]. In addition, DENV and Plasmodium possess pattern recognition molecules which induce the release of inflammatory molecules, including interleukin (IL)-1β, IL-6, IL-8 and tumor necrosis factor alpha (TNF-α) [36,37]. Because both DENV and Plasmodium are intracellular pathogens, the induction of Th1 cytokines (e.g., IL-2, IL-12 and gamma interferon [IFN]-γ) is critical for the activation of macrophages, cytotoxic T cells, and other cellular-mediated mechanisms involved in controlling replication of the pathogens. In contrast, the up-regulation of Th2 cytokines (e.g., IL-4, IL-5, and IL-13) and regulatory cytokines (e.g., IL-10 and transforming growth factor [TGF]-β1) is associated with immune escape and disease progression following DENV or Plasmodium infection [38]. Despite the need for the activation of inflammatory responses during the development of protective immunity, activation of these pathways may also result in tissue damage and severe disease [35,36,39]. Several of the pro-inflammatory cytokines induced by DENV or Plasmodium, such as IL-6, TNF-α, and IL-1β, are important mediators of acute phase proteins, as well as mechanisms of endothelial cell activation and secretion of chemokines. These molecules contribute to increased vascular permeability during severe dengue or are associated with sequestration of parasitized red blood cells during severe malaria, particularly cerebral malaria. Conversely, the induction of regulatory cytokines such as IL-10 has been involved in protection from severe disease in both dengue and malaria. Thus, the modulation of the inflammatory response following DENV or Plasmodium infection likely influences the onset of symptoms and disease outcome. Because DENV and Plasmodium both induce a type-1 immune response, co-infection theoretically could result in either enhanced pathogen clearance or increased immunopathogenesis. Given the overlapping endemicities of these two infections, it is important to understand the immune interactions that occur during co-infection, as they may lead to improved management of such cases or the identification of biomarkers to better identify coinfected cases.
To characterize symptoms and immune marker secretion profile associated with Plasmodium/DENV co-infection, we conducted an exploratory, retrospective comparative study in which we identified 17 cases with Plasmodium/DENV co-infection and matched Plasmodium and DENV mono-infected controls. These cases were identified through a febrile surveillance cohort conducted from 2002 to 2011 in Iquitos and Puerto Maldonado in the Peruvian Amazon Basin. .0010 were also reviewed by the Peruvian Ministry of Health. After being informed of the study, written consent was obtained from all participating adults or a parent or legal guardian of subjects 5-17 years old. Assent was also obtained from subjects between ages 8-17 years old. Protocol 2012.0002 covered immune marker testing of the specimens collected in the two aforementioned protocols.

Patient recruitment
From January 2002 to March 2011, subjects with an undifferentiated febrile illness were recruited at outpatient clinics and screened for the presence of an arboviral infection, as described previously [40]. This population is almost entirely formed by Peruvian citizens of mestizo origin. For this project, we only included local residents eligible for healthcare in the Peruvian cities of Iquitos and Puerto Maldonado because the Peruvian Ministry of Health often screened febrile subjects for malaria in addition to our arboviral assays to detect infection with DENV. Because the Ministry of Health's malaria screening was only incorporated into the febrile surveillance during certain time periods, samples-both mono-infected and co-infected -were only selected from these time periods. Inclusion criteria for febrile surveillance were: 1) oral/tympanic temperature > 38.0°C (or axillary > 37.5°C) or report of fever starting within the previous 5 days, 2) no obvious source of infection such as cellulitis, dental abscess, or urinary tract infection, and 3) age 5 years. Subjects infected with two serotypes of DENV (e.g., both DENV-1 and DENV-3) or two species of Plasmodium (e.g., both Plasmodium vivax and Plasmodium falciparum) were excluded from the analysis.

Collection of symptom data
Each site used a standardized questionnaire which collected the demographic and symptom information assessed in this study. In addition, all participating healthcare personnel received standardized training in data collection. We assessed 19 symptoms in our analysis: abdominal pain, nausea, vomiting, diarrhea, rhinorrhea, cough, expectoration, epistaxis, bleeding gums, hematuria, hematochezia, prostration, asthenia, chills, malaise, arthralgia, myalgia, eye pain, and headache.

Laboratory determination of DENV and Plasmodium infection
Both a 10 ml serum and a 2 ml whole blood sample were collected from febrile subjects. Microscopy was the primary method of identifying Plasmodium and culture was the primary method of identifying DENV, but DENV and Plasmodium-specific polymerase chain reaction (PCR) were used to augment these modalities when sufficient sample was available, as will be discussed later in this paragraph. A portion of the whole blood sample was used for malaria microscopy, performed at the site of collection by the Peruvian Ministry of Health using thick and thin smears according to national guidelines [41]. The remainder of the whole blood sample and the serum sample were stored at -80°C until transfer to NAMRU-6 on dry ice. At NAMRU-6, if sufficient quantity of whole blood existed, the infecting Plasmodium species was confirmed by PCR [42] on samples deemed positive by smear; Plasmodium PCR was also used to further rule out the presence of Plasmodium in smear-negative, DENV-positive specimens. To assess for the presence of DENV, serum was inoculated on C6/36 and Vero cell cultures, as previously described [40]. After ten days or upon appearance of cytopathic effect, immunofluorescent antibodies to a wide range of endemic arboviruses, including the four DENV serotypes, were added. PCR for DENV was also performed on a subset of samples, as previously described [40]. Primary versus secondary DENV infection was determined using a ratio of enzyme linked immunosorbent assay (ELISA) titers for IgM and IgG, similar to an approach used by others [43].

Classification/selection of co-infected and mono-infected cases
A subject was determined to be mono-infected with Plasmodium if they had a positive malaria smear and a negative viral culture and ELISA IgM for DENV. A subject was determined to be mono-infected with DENV if they had a positive culture or PCR for DENV and negative malaria microscopy (and PCR, if sufficient quantity of whole blood existed). A subject was determined to be co-infected if both a DENV assay (culture or PCR) and malaria microscopy were positive.
First, we identified all co-infections from the two study sites in the project database. We then selected three DENV mono-infected subjects and two to three Plasmodium monoinfected subjects for each co-infected subject (two co-infected subjects did not have Plasmodium mono-infected controls) to compare symptom frequency and levels of immune markers. All mono-infected control subjects were individually matched to a co-infected case by recruitment site, time of enrollment (±3 months), gender, and DENV serotype. Pregnant subjects were excluded from analysis due to the difficulty in finding adequately matched control subjects.

Measurement of immune markers
Serum immune marker levels were measured using the xMAP platform (Luminex; Austin, TX), following the manufacturer's recommendations. A 10-point calibration curve was used instead of an 8-point calibration curve to improve performance at the lower range of detection. The reactions were detected using the Bio-plex 100 (Bio-Rad Laboratories; Hercules, CA) system, and cytokine, chemokine, and growth factor concentrations were calculated using Bioplex Manager Software (Bio-Rad Laboratories). The limit of detection (LOD) was defined as the lowest calibrator value at which the coefficient of variance of concentration was less than 25% and recovery of calibrator was within 20% of the expected value. All values below the LOD were considered undetectable and assigned a value equal to half the plate-specific LOD. TGF-β1 was detected using a commercially available ELISA assay (R&D Systems; Minneapolis, MN). The 18 immune markers tested belonged to the following categories: inflammatory acute response cytokines (IL-1β, IL-6, IL-8, TNF-α); chemokines and growth factors (IL-2, IL-7, granulocyte colony stimulating factor [G-CSF], monocyte chemoattractant protein [MCP]-1/ CCL2, macrophage inflammatory protein [MIP]-1α/CCL3); inflammatory Th1 cytokines (IL-12, IFN-γ); inflammatory Th2 cytokines (IL-4, IL-5, IL-13); inflammatory Th17 cytokines (IL-17A, granulocyte macrophage colony stimulating factor [GM-CSF]); and regulatory cytokines (IL-10, TGF-β1). For IL-12, the assay detected the p70 protein; therefore, our assay was specific for IL-12 and did not cross-react with IL-23. All samples were tested in the same batch under the same conditions. Immune markers with values undetectable in more than 75% of the samples were not included in the analysis.

Statistical analysis
Symptom frequencies among the three groups were compared using conditional logistic regression, using a separate regression to compare the frequency of each individual symptom between co-and mono-infections. The concentrations of immune markers were compared among the three groups using the Kruskal-Wallis test, a non-parametric method that is robust and less affected by outliers. Log-transformed cytokine concentrations were compared between the three groups using random effects linear regression, again with a separate regression for each immune marker. Age and number of days with fever, and the influence of other covariates not matched for in the study design were adjusted in the regression analyses. The sample size available had 80% power to detect a difference of 0.8 standard deviations in immune marker levels between co-and mono-infected individuals.

Results
Seventeen co-infected subjects (14 P. vivax and 3 P. falciparum; 2 DENV-1 and 15 DENV-3) were identified and included in the study (Table 2). Plasmodium infection was initially identified with microscopy in all subjects and further confirmed with PCR in five (adequate sample was not available to perform confirmatory PCR on the others). DENV infection was established with culture in 15 subjects and with PCR in two subjects. Using the aforementioned matching criteria, 51 DENV mono-infected subjects and 44 Plasmodium mono-infected subjects were selected for comparison. The age distribution was similar among the three groups. The frequencies of the DENV serotypes were similar between the co-infected and DENV mono-infected subjects and the frequencies of the Plasmodium species were similar between the co-infected and Plasmodium mono-infected subjects. The average number of days with symptoms was lower in DENV mono-infected subjects compared to Plasmodium monoinfected or co-infected subjects (p<0.05; Table 2).
Only two symptoms were significantly different (p<0.05) upon comparison among the three groups: 1) more frequent myalgia in DENV mono-infected vs Plasmodium monoinfected subjects, and 2) more frequent cough in Plasmodium mono-infected vs DENV monoinfected subjects (Table 3). These differences remained significant after adjusting for age and days since symptoms onset. No symptoms were significantly more frequent in co-infection cases. Only two subjects reported epistaxis, one in the co-infected group and one in the DENV mono-infected group. No subject reported any of the other hemorrhagic manifestations-gum bleeding, hematochezia, or hematuria.
To examine the immune profile of individuals with co-infection, we compared the concentration of 15 immune markers (IL-1β, IL-6, IL-8, TNF-α, IL-7, G-CSF, MCP-1/CCL2, MIP-1-α/CCL3, IL-12, IFN-γ, IL-13, IL-17A, GM-CSF, IL-10, and TGF-β1) among the three groups using Kruskal-Wallis ANOVA. Three other immune markers (IL-2, IL-5, and GM-CSF) were excluded from the analysis given their high frequency of measurements falling below the LOD. The Plasmodium mono-infected group had significantly (p<0.05) higher concentrations for the majority of the immune markers tested compared to the DENV mono-infected or the coinfected groups (Fig 1). The consistently higher levels in Plasmodium mono-infected subjects were not particular to a specific type of immune marker, as significant differences were found in the inflammatory acute response (IL-1β, IL-6), chemokine and growth factor (G-CSF), Th1 (IL-12, IFN-γ), Th2 (IL-4), Th17 (IL-17A), and the regulatory cytokine IL-10. In contrast, the concentration of TGFβ1, a regulatory cytokine, was significantly higher in the DENV monoinfected group compared to either the Plasmodium mono-infected or co-infected groups. DENV mono-infected subjects and co-infected subjects had similar levels for all other immune markers except IL-4 and IL-7, which were both higher in the DENV mono-infected group compared to the co-infected group. IL-4 was the only immune marker with statistically significant differences across the three groups, with the Plasmodium mono-infected group having the highest and the co-infected group the lowest level (Fig 1). No immune marker was significantly higher in the co-infected group compared to either of the mono-infected groups.
To assess for the influence of other covariates which were not matched for in the study design, we fitted a random effects regression model for matched group analysis, adjusting for age and number of days with fever. This analysis confirmed the above results in that the Plasmodium mono-infected subjects had significantly higher serum concentrations of IL-4, IL-6, IL-10, IL-12, IL-13, IL-17A, MIP-1α/CCL3, and IFN-γ (p < 0.05) than either the co-infected or the DENV mono-infected subjects (Table 4) even when adjusting for age and the number of days. Additionally, TGF-β1 was the only immune marker significantly higher in the DENV mono-infected subjects than in the Plasmodium mono-infected subjects.

Discussion
This study adds Peru to other South American countries reporting Plasmodium/DENV coinfection in the Amazon region. Our results revealed a strikingly similar inflammatory profile between subjects with co-infection and DENV mono-infection, in sharp contrast to the higher levels of immune markers in those with Plasmodium mono-infection. On the other hand, most symptoms occurred with similar frequencies between the three groups, although the study was most likely underpowered to detect differences. Because the Ministry of Health's malaria microscopy samples were incorporated into this febrile surveillance protocol only during certain time periods over the nine-year study, we did not calculate an incidence of Plasmodium/ DENV co-infection for this timeframe. In addition to this investigation, only five studies of Plasmodium/DENV co-infection have compared co-infected subjects to those with mono-infection [10,17,21,27,30]. A study from Pakistan looking at 19 clinical features found only one to be different in co-infected patients, jaundice, and this occurred less often in the co-infected group than either Plasmodium or DENV mono-infection [30]. A group from India followed 27 co-infected, 102 Plasmodium mono-infected, and 340 DENV mono-infected subjects and compared symptoms and routine laboratory values, including liver function tests and hematologic parameters [21]. P. falciparum was the most prevalent malaria species in this study, although they observed trends similar to ours where P. vivax predominated. Specifically: 1) similarities between Plasmodium/DENV coinfection and DENV mono-infection; 2) differences between Plasmodium/DENV co-infection Table 3. Symptom frequency in subjects with co-infection, dengue virus (DENV) mono-infection, and Plasmodium mono-infection.

Symptoms
Co-infection n/total (%) DENV n/total (%) Plasmodium n/total (%) p value* and Plasmodium mono-infection; and 3) no apparent worsened disease during infection with both pathogens. Conversely, the largest comparison study of co-infected subjects, which described 104 co-infected patients in French Guiana, showed that hypotension,  ) for all immune markers except TGF-β1, which was measured by an R&D Systems ELISA kit. Three immune markers were excluded (GM-CSF, IL-5, and IL-2) from this analysis because more than 25% of the values fell below the limit of detection of the assays. Horizontal bars represent group median and inter-quartile range. Red dotted lines show lower limit of detection for each immune marker. Units on the y-axis are log 10 transformed scales of analyte concentration in ng/mL for all immune markers except for TGF-β1, which is represented in μg/mL. * p<0.05 for group comparison by the Kruskal-Wallis test followed by post-hoc Dunn's pair-wise comparison.
doi:10.1371/journal.pntd.0004646.g001 thrombocytopenia, and anemia were more common in co-infection than in mono-infection with either pathogen alone [17]. Similarly, a study from Brazil showed more bleeding, hepatomegaly, and jaundice in the co-infected group [27]. Furthermore, a study from Pakistan contrasted 26 co-infected subjects with 52 DENV mono-infected subjects and found that hematocrit, hemoglobin, and platelet levels were all lower in the co-infected group although leukopenia was more common in those with DENV mono-infection [10]. However, symptoms were not compared, tests of statistical significance were not performed, and no attempt was made to match or adjust the comparator groups for other covariates. The differences in the findings among these five studies and ours-with three demonstrating and three not demonstrating worse disease in co-infected subjects-may be due to differences in DENV serotypes, Plasmodium species, or outcomes evaluated. While tempting to hypothesize that Plasmodium/DENV co-infection would routinely result in more severe outcomes, as seen in the French Guiana study, most co-infection case reports do not describe severe or fatal outcomes ( Table 1). The majority of co-infection cases in the literature occurred in regions, similar to ours, with a predominance of P. vivax, a malaria species less associated with severe outcomes than P. falciparum [44]. All case reports describing fatal or near-fatal Plasmodium/DENV co-infection involved P. falciparum, including those from West Africa [6], Indonesia [24], East Timor [9], Pakistan [7], and India [16]. Nevertheless, severe manifestations are still possible when Plasmodium/DENV co-infection occurs with P. vivax, as demonstrated in the larger studies from French Guiana and Brazil [17,27]. Both malaria and dengue have been associated with a strong activation of acute phase response (IL-6, TNF-α, and IL1-β) and Th1 cytokines (IFN-γ and IL-12). Therefore, we hypothesized that co-infection may result in synergistic activation of these pathways and overall higher levels of pro-inflammatory immune markers in the serum. However, our comprehensive characterization of multiple cytokines, chemokines, and growth factors using a highly sensitive multiplex platform revealed that Plasmodium mono-infected subjects had higher levels of many of the immune markers evaluated compared to DENV mono-infected subjects even after matching and adjusting for other covariates. The lower levels of inflammation in the co-infected group were, in fact, similar to DENV mono-infected subjects. Consistent with our results, P. falciparum infection has been associated with a greater activation of inflammatory response compared with DENV infection, including higher levels of IFN-γ and IL-10 [45] and, in some cases, P. vivax may have an even more robust inflammatory response than P. falciparum [46].
The regulatory marker TGF-β1 was the only immune marker with higher levels in DENV mono-infected individuals compared to Plasmodium mono-infected and co-infected subjects. Prior studies of TGF-β1 in DENV mono-infection have demonstrated a positive correlation with symptomatic disease. A group from French Polynesia showed that TGF-β1 levels were highest in those with dengue hemorrhagic fever versus dengue fever; likewise, higher levels were found in those with dengue fever than healthy controls [47]. A study of subjects with dengue in India similarly showed a correlation of TGF-β1 levels with severity and also length of disease [48]. In our study and others, TGF-β1 levels in malaria show an opposite trend compared with dengue. TGF-β1 levels were higher in healthy Tanzanian controls compared to those with P. falciparum malaria. In the same study, TGF-β1 levels were inversely correlated with malaria severity [49]. Parasite density in children from Gabon with P. falciparum malaria was also shown to be inversely related to TGF-β1 levels and highest in those without malaria [50]. Our findings are consistent with these four studies, which together demonstrate high TGF-β1 levels in symptomatic dengue and low TGF-β1 levels in symptomatic malaria.
Following infection with DENV, viremia first peaks and then becomes undetectable over a timespan usually lasting less than three weeks. Chronic infection is not a characteristic of dengue. Asymptomatic, long-lasting Plasmodium infections, however, are common both in P. vivax and P. falciparum malaria [51]. In the case of P. vivax, a clinically dormant hypnozoite form exists that may reactivate months or years later during a physiological stressor, such as DENV infection. The scenario of acute, symptomatic DENV infection occurring in a person with clinically silent Plasmodium infection would explain why the immune marker profile of our co-infected subjects nearly always resembled DENV mono-infected subjects. In contrast, our subjects with Plasmodium mono-infection most likely presented with fever directly attributable to malaria and therefore displayed a vastly different immune marker response from the other two groups. This same reasoning would account for why Plasmodium/DENV co-infection did not result in a higher frequency of symptoms than mono-infection with DENV. Further studies in humans or animal models are needed to better understand the effect of DENV infection on the course of Plasmodium infection or vice versa.
A few caveats are worth noting. First, given the retrospective nature of our study, we measured soluble immune marker levels in the serum which are not pathogen-specific, so it is possible that other infections or underlying medical conditions may have confounded our findings. However, we individually matched mono-infected subjects to every co-infected subject in terms of gender, enrollment site, and period (±3 months). In addition, we performed immune marker testing on all samples at the same time in order to avoid batch variability.
Second, due to the design of the studies, we did not collect clinical laboratory data, such as platelet levels, creatinine, liver enzymes, or hematocrit, which can be used to stratify disease severity and has been utilized to differentiate between Plasmodium and DENV mono-infection in other studies [52]. Furthermore, we only assessed our subjects at one point in time, which could have missed evolution to a more severe disease state, such as dengue hemorrhagic fever, which classically follows an initial, less severe dengue fever stage. However, dengue hemorrhagic fever was virtually non-existent in Iquitos and Puerto Maldonado during the time of our study. Next, our study occurred during a timeframe when DENV-3 and P. vivax were prominent in the study regions and infection with P. falciparum was not associated with severe outcomes, so our results may not be generalizable to areas in the world with different DENV serotypes or a predominance of P. falciparum malaria. Finally, our sample size may have been too small to detect differences in most symptoms; however, large and statistically significant differences in immune marker levels were still able to be observed, even though the study may have been underpowered to detect more nuanced differences.
Our results verify the existence of Plasmodium/DENV co-infection in the Amazon region of Peru. Therefore, diagnosis of one disease in a febrile patient in this area does not preclude infection with the other. This realization is significant because malaria can be treated with prompt administration of anti-malarial medications, which could be mistakenly withheld if only DENV were presumed in a co-infected patient. Although effective anti-viral treatment is not available for dengue disease, early identification is crucial for the supportive management of severe cases [53]. For the time period of our study, co-infection did not yield more symptoms or higher immune marker levels than mono-infection alone. While this is momentarily reassuring, the recent discovery of more life-threatening forms of malaria [54] and DENV [55] in the region emphasize the future importance of not overlooking either of these potentially fatal but treatable diseases.

Disclaimers
The opinions expressed in this work are those of the authors and do not reflect the official policy or position of the Department of the Navy, Department of Defense, U.S. Centers for Disease Control and Prevention, or U.S. Government.
Some of the authors are military service members or employees of the U.S. Government. This work was prepared as part of their official duties. Title 17 U.S.C. §105 provides that 'Copyright protection under this title is not available for any work of the United States Government.' Title 17 U.S.C. §101 defines a U.S. Government work as a work prepared by a military service member or employee of the U.S. Government as part of that person's official duties.